That factoid about the heaviest drinking 10% is probably wrong, by the way.

Basically, it comes out of survey data about how much alcohol people consume. But when you take the actual results from the survey and multiply it by the population of the US, you only get about half as much alcohol consumption as there are sales of alcohol.

So the guy who ran the survey just multiplied all the numbers by two. People who said they had 5 drinks per night, he assumes they have 10 drinks per night. In contrast, the people who said they have 1 drink per month, he assumes 2 drinks per month.

That's pretty dubious. It seems like it's a LOT more plausible that if you have 3 drinks per month, you'd report just one, than if you report 5 drinks per night, actually you take 10.

Reference: http://vinepair.com/booze-news/10-americans-consume-10-drink...

Wow, thank you. That really changes the initial result.

I remember being shocked by those numbers. 10 per night is high-impact even with high tolerance, and totally out of the question for lots of people. For 5% of Americans to drink more than that would be wild, since you've largely slashed your potential demographic to good-sized men.

And the reverse explanation, where the low end is particularly bad at estimating, would match with everything else people estimate. If you have a beer after work and a second with dinner, you can probably say "two a day" with confidence. If you have a glass of wine sometimes, out with friends, but not often, it's not terribly easy to guess whether that's two a month or one every two months.

I sort of see why he used his estimate, because you have to bring up the low end hugely to account for so much absence, but it seems unjustifiable to assume that the heaviest drinkers could even survive having their numbers doubled.

Yeah, I think that probably a huge chunk of the difference is just that actually a lot of alcohol purchased is disposed of unconsumed.

Tie that with a modest increase in the lower-end reporters, and perhaps a modest decrease in the highest-end reporters (are there people who said they average 5 drinks a night who really average 4 because in fact their drinking is variable per night and averaging across a week is more mental record keeping + math than they bothered with when answering a survey?), and you get something that strikes me as more congruent with basically all other facts that I've observed about the world.

Now, that probably does still mean that the top decile accounts for an astonishing percentage of overall alcohol consumption, on a relative basis -- just the absolute amount is less incredible.

The 'failure to consume' part didn't even occur to me. Between half-emptied drinks that go bad and liquor cabinets which have been stocked but barely used for decades, it can't be a small amount.

And yeah - I don't doubt it's a power law thing, pretty much all addictive substances are. I imagine the 95th percentile of use for a lot of drugs would be lethal for a first-time user. But the raw dose claimed for alcohol here was bizarre, and I'm glad to have it explained.

> Freemium devs can't fly in wealthy players and customize their games, for instance.

I wish I could remember the source but I'm sure I've heard of freemium developers researching their biggest whales and then tailoring high end items to match the colours of their favourite football team.

They absolutely do stuff like that. There's a freemium game developer that lives in my neighborhood. Something crazy like 50% of their income came from just a handful of people (literally - something like five people). Naturally, the devs made the effort to reach out and figure out why those people were buying things and what the company could do to keep them buying, asking what features they'd like to see and what models they liked.

Do most capitalistic ventures rely on whales? Or is it limited only to ones that feed off addictions?

Addictive ones only, basically.

There's two things restraining other businesses from profiting like this.

First, lots are basically fixed-consumption - if you buy a vacuum cleaner, there's effectively no chance of selling you another one next week. Across all your customers, you have an effectively flat purchase rate. So all you can do there is improve margins and reach unserved customers (and maybe pursue planned obsolescence).

Second, most unbounded-consumption businesses have normal distributions of consumption. People who buy string cheese maybe eat an average of five a week, with a low-SD normal distribution around that. So you market your cheese sticks to your prototypical customer to maximize profit.

Any power-law business sells to whales. That means drugs and gambling, but I should also observe that enterprise sales count. If Proctor and Gamble will pay 10,000X as much for your security consulting as a local pizza shop, you're selling to whales.

Some games do target individual whales, at least according to one article on gamasutra.